arXiv:2610.12454
On PPT entanglement distillation
Ludovico Lami
Abstract
We study entanglement distillation under operations that remain completely positive under partial-transpose conjugation, a.k.a. PPT channels, in the standard quantum Shannon theory regime of asymptotically vanishing but non-zero error. Our main results are: (a) a regularised formula for the PPT distillable entanglement in terms of a measured-relative-entropy-like quantity; and (b) two different single-letter converses, one based on operator quadratic forms and the other on Hirschman's strengthening of the Hadamard three-line theorem. As an immediate application of (b), we show that the PPT distillable entanglement can be strictly smaller than the regularised Rains bound, thereby resolving an open problem in the theory of entanglement manipulation [Regula et al., NJP 21:103017, 2019]. A gap appears already for $3\times 3$ Werner states: at antisymmetric weight $25/26$, a certified upper bound of $0.62107$ ebits lies below the (regularised) Rains bound, which equals $\frac{25}{26} \log_2 5 - \log_2 3 \approx 0.64766$ ebits. On a different note, (a) implies a faithful lower bound on the PPT distillable entanglement in terms of the entanglement negativity, which provides a quantitative counterpart to the qualitatively known fact that any NPT state is PPT distillable [Eggeling et al., PRL 87:257902, 2001].
AI Summary
The paper studies entanglement distillation under completely PPT-preserving channels in the asymptotic, vanishing-error regime. Its central question is whether PPT distillable entanglement equals the regularised Rains bound. It establishes $E_{d,\mathrm{PPT}}(\rho)=L^\infty(\rho)$, where $L(\rho)=\sup_{(E_j)}\sum_j p_j\log_2[p_j/\|E_j^\Gamma\|_\infty]$ and $p_j=\operatorname{Tr}(\rho E_j)$. Achievability uses repeated measurements and classical typicality; the reverse inequality pulls back the Bell test of a distillation protocol. An auxiliary measurement functional yields the faithful lower bound $E_{d,\mathrm{PPT}}(\rho)\geq1-h_2(1/2+N(\rho)/(d+1))$, with $d=\min(d_A,d_B)$, quantitatively strengthening the known distillability of every NPT state under PPT operations. Two converse constructions use quadratic forms on operator spaces and analytic matrix families with Hirschman’s three-line inequality, respectively. Both provide exponential strong converses and also bound LOCC distillation. For the two-qutrit Werner state with antisymmetric weight $25/26$, they establish $0.5836\leq E_{d,\mathrm{PPT}}\leq0.62107$, strictly below $R^\infty=(25/26)\log_2 5-\log_2 3\approx0.64766$. Thus equality with the regularised Rains bound fails. The lower certificate uses a feasible 48-copy POVM; the quadratic separation reduces to a two-dimensional calculation, while the tighter analytic bound uses directed-rounding quadrature. The characterisation remains regularised, with no convergence-rate guarantee. The quadratic framework requires an all-copy constraint that does not follow from one-copy validity. The numerical certificates do not establish optimal rates or optimal converse parameters. A single-letter formula, general computability, and a complete convergent family of upper bounds remain open.
The main results and techniques presented here have been derived almost entirely by ChatGPT Astra.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 31+7 pages, 1 figure